<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Prashant Pandey</title>
    <link>https://pandeyps.in/blog.html</link>
    <atom:link href="https://pandeyps.in/feed.xml" rel="self" type="application/rss+xml" />
    <description>Blog posts by Prashant Pandey</description>
    
    <item>
      <title>world models and using them to dream drive</title>
      <link>https://pandeyps.in/blog/drive-wm/</link>
      <guid isPermaLink="true">https://pandeyps.in/blog/drive-wm/</guid>
      <pubDate>Tue, 25 Aug 2026 00:00:00 +0000</pubDate>
      <description>You can find the full code here: https://github.com/pandey-ps/world-models/tree/master/drive (it is well commented to explain about the environment of the model).

</description>
    </item>
    
    <item>
      <title>Building LLMs from Scratch</title>
      <link>https://pandeyps.in/blog/building-llms-from-scratch/</link>
      <guid isPermaLink="true">https://pandeyps.in/blog/building-llms-from-scratch/</guid>
      <pubDate>Mon, 30 Mar 2026 00:00:00 +0000</pubDate>
      <description>DeepSeekV3: You can find the full code here: https://github.com/pandey-ps/papers/tree/main/deepseekv3

</description>
    </item>
    
    <item>
      <title>Ray Tracer in 99 Lines of Python</title>
      <link>https://pandeyps.in/blog/raytracer/</link>
      <guid isPermaLink="true">https://pandeyps.in/blog/raytracer/</guid>
      <pubDate>Sat, 28 Feb 2026 00:00:00 +0000</pubDate>
      <description>You can find the full code here: https://github.com/pandey-ps/raytracer

</description>
    </item>
    
    <item>
      <title>Dynamic Resolution Vision Transformer: Faster Inference Without Extra Training</title>
      <link>https://pandeyps.in/blog/dvit/</link>
      <guid isPermaLink="true">https://pandeyps.in/blog/dvit/</guid>
      <pubDate>Sat, 10 Jan 2026 00:00:00 +0000</pubDate>
      <description>Vision Transformers run at a fixed image resolution. Easy images still go through the full high resolution pipeline, wasting time and compute.

</description>
    </item>
    
    <item>
      <title>Implementing DeepSeek-OCR</title>
      <link>https://pandeyps.in/blog/deepseek-ocr/</link>
      <guid isPermaLink="true">https://pandeyps.in/blog/deepseek-ocr/</guid>
      <pubDate>Thu, 20 Nov 2025 00:00:00 +0000</pubDate>
      <description>DeepSeek recently released DeepSeek-OCR, the research paper of it focuses on vision text compression, the model can decode thousands of text tokens from few hundred vision tokens. I wanted to test this, so I set up a small Colab pipeline to see how well it works.

</description>
    </item>
    
  </channel>
</rss>
